The purpose of this study is to describe the relationship between sleeping and waking states during the preterm period and the later development of premature infants. The developmental patterns of sleep-wake states and state-related behaviors during the preterm period will be studied and related to multiple dimensions of development at age 3: social competence, cognitive abilities, sleep patterns, health status, visual-motor coordination, language skills, and gross motor abilities. Since the social environment has a major impact on infant development, the quality of this environment will also be measured at age 3. Subjects for this study will be two cohorts of 35 premature infants who will be recruited from the NICU as soon as their medical conditions are not critical. Infants in the first cohort are currently being studied in a project on sleep-wake states during the preterm period. They will be followed at age 3. Infants in the second cohort will be studied during the preterm period and at age 3. During the preterm period, infants will be observed for 4 hours at weekly intervals until reaching term age or until hospital discharge. A minimum of two observations will be obtained on each subject. During the observation, the occurrence of 10 states within sleeping and waking and 16 infant behaviors will be recorded every 10 seconds with an event recorder. In addition, the infant's respiration will be recorded so that state-related respiratory patterns and apneic pauses can be identified. The severity of the infant's illness will be rated on a daily basis. AT 2 years and 9 months post-term (about 3 years chronological) assessments of the child's developmental status will be made: two observations of child social and play behaviors during 2-hour periods in the home, parental recording of sleeping and eating patterns, the HOME, NCAST Teaching Scales, Stanford-Binet Intelligence test, and the PEET. Data analysis will concentrate on describing the development of sleep-wake states and state-related behaviors during the preterm period and on relating early behavioral patterns to later developmental outcome. The second cohort will be used to verify relations found in the first cohort.